What happens week by week
Point at any chart to read the values for that week; the model diagram below follows the same week. Dashed grey lines are the saved baseline.
Demand and production
Cases per week. The bullwhip is production swinging wider than the demand that drives it.
Stock
Cases on hand that can still be sold.
Service
Share of each week's demand that reached customers.
Cumulative profit
The plant's revenue minus every cost, from week 0.
Workforce
People on the payroll; capacity follows them.
Waste
Cases lost so far: batches rejected in QA, and stock written off because too little shelf life remained.
Decision lab
Compare saved scenarios side by side, see the range of outcomes over many demand futures, find which inputs matter most, and let the optimiser search the planning policy. Every option is tried on the same random futures.
Save the current scenario, change things and save again: every saved scenario is rerun here side by side, with the best value in each row highlighted. Saved scenarios stay in this browser.
No saved scenarios yet: the table shows only the current one.
The same scenario in many random demand futures, and the baseline in the very same futures. The shaded band holds the middle 80% of futures; the line is the median.
Service across futures
Production across futures
Which inputs matter most? Each input is moved 20% down and 20% up, one at a time, and the bars show how far the chosen result moves. The sweep below runs one input across its whole range.
Searches 540 planning policies (target cover, time to correct stock, supply-line weight, forecast smoothing, overtime and expedited QA) over the same demand futures, and ranks those that reach the service target by average profit. Everything else stays as in the current scenario. Apply any result, or click a point, to explore it.
The model
Boxes are stocks, valves are flows and circles are what the planners see. The numbers are for the week shown, which follows your pointer on the charts.
Every equation, with its value in that week
How to read it: what the results mean
Fill rate is the share of market demand that reached customers. In food retail the target is usually 98–99.5%, so a fraction of a point is a real shortage. Service slips before the stock runs out, because stock is spread over many pack sizes, flavours and customers.
Bullwhip is how much more production varies than demand (standard deviation of production ÷ standard deviation of demand). Above 1, the plant amplifies what the market does: it is chasing its own stock gap, not just demand. With a distributor, the same ratio for its orders shows where the amplification starts.
Waste is batches rejected in QA plus stock written off because too little shelf life remained for customers to accept it. With a long shelf life it is almost all rejections; with a short one, an overshoot after a surge becomes expired stock.
Profit is the plant's view: revenue minus materials, the workforce (paid whether busy or not), hiring, overtime, holding, disposal, expediting and a penalty for each case not supplied. The figures are illustrative, chosen for a mid-priced branded food product.
The baseline is a saved run for comparison. It uses the same random demand as the current run, so every difference is due to what you changed, not to luck. New demand future draws another random future for both.
What the model leaves out
- One product family, one plant, at most one distributor and one ingredient supplier. Several markets, other products competing for the same line, and allocation between markets are not modelled.
- Campaigns are averaged: their lost capacity, delay and cycle stock are modelled, but not individual batches and when each is made.
- The distributor's stock is not aged; shelf life is tracked at the plant only.
- Nobody is laid off: the workforce shrinks only as people leave.
- Costs and prices are illustrative, not a quotation.